Agentic LXP · Layer 2 · Agent stack

AI agents: a team that helps out in everyday learning

Four specialized AI agents take on tasks in everyday learning — unlike a chatbot, with tools and in the context of your platform.

02 The challenge

More learning needs, the same number of hands

Skills change faster than learning offerings can be built by hand. And a chatbot answers questions but does not get any work done.

?
01 Problem

Gaps stay invisible

Which skills the team is missing often only becomes clear when it is too late.

NEW MODULEin progress … week 3
02 Problem

Content is created too slowly

By the time a suitable learning module is ready, the need has already shifted.

INBOX12open questionsQuestion about lesson 2 …Question about lesson 3 …Question about lesson 4 …
03 Problem

Learners are left alone

Questions about the course land in the inbox, and administration takes up time.

03 How it works

The relay: four agents, one task

Scroll through five chapters. Click a chapter to see its result.

Your agent team
Four AI agents side by side: skill agent, content agent, tutor agent and admin agent, above them the “Choose model” bar and the “Hosted in the EU” badge.MODELGPTClaudeMistralOpen sourceEU hostingContextPerceptionMemoryToolsAction✓ detected✓ created✓ answered✓ recordedTASKGap: fire safetyDialog in the course✓ approvedSkill agentSkillsContent agentLearning contentTutor agentSupportAdmin agentAdministration
In action

01 · Team

YOUR AGENT TEAM
Skill agentAI AnnotationsDetects skill gaps along skill taxonomies such as ESCO and tags content with skills.Uses: ESCO taxonomy, skill mapping
Content agentAI FellowCreates, curates and adapts learning content, for example an xAPI module from a PDF.Uses: xAPI Package Builder, WCAG Auditor, brand style guide
Tutor agentAI TutorSupports learners in dialog, in the context of the course: answering questions, summarizing, plain language, translating.Uses: content of your platform
Admin agentAdministration & ReportingHandles administration and reporting: invitations, users, roles, reports.Uses: REST API, learning analytics

02 · Anatomy

CHATBOT

Create a fire safety module for my team.

Sure! Here are five fire safety tips …

Result: a text
content-agent

$ agent “Create a fire safety module”

› context · courses, profiles, learning activities

› memory · previous steps

› tool xapi-package-builder

✓ action · fire-safety.zip created

Result: a completed task

03 · Relay

relay

› skill-agent · gap detected: fire safety (ESCO)

› content-agent · module created: fire-safety.zip

› tutor-agent · question answered, lesson 3

✓ admin-agent · result recorded

LEARNER 1 · SALES TEAM
Fire safety
Gap
Fire safety training8 lessons · xAPI
new
Tutor agent answers in lesson 3
Progress
completed ✓

04 · Support

COURSE · FIRE SAFETY TRAINING · LESSON 3

Tuesday, 9:14 pm

What do I do if I notice a fire?

Stay calm, report the fire and warn people in danger. The steps are in lesson 3.

Source: lesson 3 · section 2
SummarizePlain languageTranslate

05 · Control

APPROVAL
Action awaiting approval

Admin agent: send invitations to the sales team

RejectApprove

✓ Approved by the L&D team

log

✓ 21:05 · admin-agent · approved · logged

Model
GPTClaudeMistralOpen source
Hosted in the EU

01 · Team

Four agents, clearly divided tasks

Skill agent, content agent, tutor agent and admin agent. Each has a task and the tools for it.

What happens
The four agents appear one after another, each with its task and its tools.
What you get
For every task, an agent with the right tools.

02 · Anatomy

A chatbot answers, an agent acts

Context, perception, memory, tools, action: an agent understands the situation, remembers, reaches for the right tools and completes the task.

What happens
The same request: the chatbot writes tips, the content agent creates the module.
What you get
The result is a completed task, not just a text.

03 · Relay

Closing a skill gap

The skill agent finds the gap, the content agent creates the matching module, the tutor agent supports the learner, the admin agent records the result.

What happens
A task moves from agent to agent, each completes its step.
What you get
From detection to recording, every step has its agent.

04 · Support

Questions in the evening, answers from the course

The tutor agent answers questions in the context of the course, pointing to the relevant passage — in plain language or translated on request.

What happens
A learner asks in the evening, the tutor agent answers from lesson 3.
What you get
Answers from the course content, with a source reference, in plain language or translated on request.

05 · Control

Your models, your approval

You choose the AI model, the data stays in the EU, and a person approves actions. Every action is logged.

What happens
A human approves the admin agent's action, the log records it.
What you get
You choose the model, the data stays in the EU, every action is logged.

● What you get

Work instead of answers

The agents take over the whole chain: detect the gap, create the module, support learners, record the result.

In the context of your platform

They know profiles, courses and learning activities — their suggestions fit your organization.

You stay in control

You choose the AI models, the data stays in the EU, people approve.

FAQ

Frequently asked questions about the AI agents

What do AI agents do in an Agentic LXP?

The AI agents of the Agentic LXP are a team of four specialized agents — skill agent, content agent, tutor agent and admin agent — that take on tasks in everyday learning and call the skills from layer 1 to do so. They detect skill gaps, create suitable content, support learners and handle administration.

Who are the AI agents for?

courseticket uses AI agents in its Agentic LXP that detect skill gaps, create learning content, support learners and handle administration for organizations and education providers.

What is the difference between an AI agent and a chatbot?

A chatbot answers. An AI agent has tools and context: it knows courses, profiles and learning activities, calls skills and uses them to complete tasks. To do this, it combines five capabilities: context, perception, memory, tools and action.

How does the tutor agent support learners?

It answers questions in dialog, in the context of the course and with a reference to the passage the answer comes from. On request, it summarizes, rephrases in plain language or translates. More: AI tutor.

Why AI agents, and why now?

Skill requirements are changing with AI itself. If you keep planning learning offerings by hand only, you cannot keep up. The models behind the agents are interchangeable: the team grows with the platform and with new skills, without you having to switch systems.

Which AI models are used, and where is the data stored?

The models are selectable, including GPT (also via Microsoft Azure), Claude, Mistral, Gemini and open-source models such as Llama. EU-hosted models are the default; for sensitive areas, closed or self-hosted models are available.

Do the agents replace trainers?

No. They take over recurring tasks, people keep oversight: a person approves actions, and every action is logged. For certified courses, a human supervisor remains mandatory.

Can I use the agents without the platform?

Partly. The skills the agents work with also run locally with your own AI agents, such as Claude Code or Codex. What the agents create, such as an e-learning module, is a standard xAPI package for any xAPI-capable LMS. The context from profiles, courses and learning activities and the orchestration of the agents come from the platform.

What does “Agentic LXP” mean?

An Agentic LXP is a learning platform in which specialized AI agents personalize learning paths, curate content and support learners — orchestrated by the platform instead of managed by hand. The AI agents are its middle layer. To the overview

Demo

See the AI team at work

In a demo, we show the agents at work in a learning environment like yours.